Honestly, this is pretty straight-forward stuff, and I'm surprised you're getting any argument. If a technologist cannot explain how a proposed implementation is going to provide value to the business, that project is probably better off left alone. Prettier code and better tests don't provide value. But fewer customer-facing P1/P2 tickets provides value. That prettier code and better tests leads to fewer tickets is an implementation detail.

Let's put it this way: Netflix has one of the most robust microservice architectures I know of. I promise you will find zero mention of it anywhere on their customer-facing pages. Because that is not the problem that Netflix solves for their customer; they solve a media library and streaming problem.

The business is your customer. You need to understand their problems and how you're going to fix them. Everything else is an implementation detail.

>Honestly, this is pretty straight-forward stuff, and I'm surprised you're getting any argument.

I'm not. Our industry is full of academics who've been told "this is the right way to do it, everything else is stupid" for the entirety of their education. They're academics, trained by other academics, ignoring the broader reality of why we do these things at all.

The corollary here that it is perfectly acceptable for business users to not understand significant parts of their own business operations. In fact, everybody should work around your shortcomings in understanding that "technical" stuff. As the guy with the MBA, you are the most important person in the room and everybody else is there to serve you.

You are a team. Not everyone has the same background and experiences. You would be amazed at how many people in an organization have incredible knowledge of their business ___domain and little to no knowledge of IT systems, their costs, and benefits. As an IT leader it is your responsibility to help bridge that gap, and the language of business is business not tech jargon, so I would caution against dismissing it.

Another way to think of it is as of user acquisition problem. You have to highlight the benefits in a way that is compelling to them. At the end of a day all progress/fiascos in a company comes down to consistent and effective communication.

Still doesn't work. I've done clandestine work over weekends to speed up an application, then as part of shipping it also add a split test with half the audience having a delay similar to what was in place before I fixed the problem and brought that to the table with the "business" guys, and everyone nodded their head "wow" but every sprint after that was just full of new features.

I hope you got the lesson. Please stop working over the week ends for free.

You are devaluing yourself and you are devaluing all the developers who don't want to ruin their week end at work.

I obviously didn't bring to the demonstration that I had volunteered the time, it suited my point to make it seem like it was something easy I could do as an aside... Problem is, they don't care about $ either, they care about what makes them look good, what makes their boss look good, what the sales guys are bugging them about, and it's features and visual polish and never quality or performance.

I had to break it down to one of them "Listen, we've seen that the site being faster gets us X% more conversions, if the site was down, or if the site was ugly X% of the time and cost us those conversions, you'd be breathing down my neck to fix it, why is this any different...?" reply after a few seconds of thought "You're right, but you'll never convince Leadership"

> it suited my point to make it seem like it was something easy I could do as an aside...

Thus fuelling the idea that this sort of thing will just be done as other work is happening with no time budgeted for it.

> I've done clandestine work over weekends to speed up an application

> I obviously didn't bring to the demonstration that I had volunteered the time, it suited my point to make it seem like it was something easy I could do as an aside

I understand the impulse, but in long term this leads to inflated estimates from management - and tech side has only themselves to blame. When we pretend things take shorter then they really do, dont be surprised when they learn to expect our lie to become truth.

Long term here can be as short as three months.

There is an old Google paper where they talk about discovering that users returned more often if the search results list was 10 instead of 20, since the shorter list took marginally less time to load/render for the user. If your co-workers don't believe you then I am sure they would at least love the Google finding that minor performance improvements have a positive impact. They would then support improving performance in the app and brag about how much like Google they are.

>>> There is an old Google paper

No, there isn't.

Google is showing one series of ads per page. The study showed that having more legitimate results per page decreased their revenues. Of course, it gives less room for ads and less page views.

I saw your post but didn't reply.

The work you did probably is great and do I think its above your average developer. Somewhere in the the top 10%. Though how you conducted yourself in the face of management was down right petty/folly.

Yes you did the best job in the world. You improved the performance of the serves by % and it did make management of those server's easier. Though you need to think about what you've done in the eyes of management.

Firstly you've modified the server's without permission. You've run tests out-side of work hours that could of impacted the business bottom line. For all I know (its up for debate) but your senior managers had more pressing issues on the table, and having a young code money running around outside office hours is the LAST thing they want to hear.

`It's not a simple as you may think it is.`

Some of the things that come to mind, and honestly I would of pulled you into a office to explain yourself if you're on my team.

1) Did you document all the changes you've made to the server

2) Did you back-up or store any sensitive information off-site during out of office hours

3) Did you modify critical system files that could cause another dependent system to fail

4) What ETA/Service level agreements if any did you break when bringing down the server for maintenance?

5) Did we lose any Security Certification with you modifying the server or changing any of the certificates on the server?

We're software developer, and yes when we see problem's we want to fix it to our best of our ability. Though when you're in a Business there are many thing you may not be aware of. In this situation I would honestly say you're lucky to keep your job! I hope you take it into consideration next time you want to go out-side of your duties.

I don't want to come down harsh on you, but there are times when you see a bad situation and there is nothing you can do about it. You notify the people who're higher than you of the situation and the dangers. It's then up to them to make the decision. After they've made the decision that is it. If the server blows up during the weekend, and they want you to come in and fix it (even when its there fuckup) you just take your hourly rate and x10. So if your hourly rate is $50hr and the server blows up, then your new rate is going to be $500 per hour to come in during the weekend to fix the mess.

I have a counter-perspective to provide:

Many of the points you've raised are moot outside of large corporate environments and projects.

It is likely that the commenter has root access to all of the things you mention and touches them on a daily basis. It is likely that they are not inundated with all of the heavy process you're bringing up. It is likely that all of the security concerns you raise are not even part of their current business. Working outside of set business hours is likely not a concern; overtime is likely uncompensated anyway and there probably aren't considerable office security protocols (they probably have keys and an access code for the whole office).

The points you've raised do not apply to most small development teams.

Yes, but both of these viewpoints are important, depending on the context.

It was a simple performance bug (think N+1 query type) that was trivial to implement and trivial to be confident about its behavior vs the pathological implementation, and I used the normal channels for shipping it, it passed QA.

The only "clandestine" part is that I added the split test, and did some work outside of the set of features that was scheduled.

We had no policies around SLA, developing off-site, security or restricting sensitive information (think potential FERPA/HIPPA violation) which are yet other "non-features" that I tried to get them to adopt.

I implemented this entirely above-board, with more precautions and consideration than were taken by other developers on that team for the day-to-day.

Agreed, but don't take this as "don't work hard," merely as "don't donate time to your employer." There are far more worthy causes out there.

The core conflict is this:

The engineer is trying to optimize/normalize his free time by acknowledging quality issues, the management is trying to optimize their income by sweeping them under the rug.

The engineer is ultimately liable for the problem. Quality is insured with his reputation and free time.

The business guys almost always leave at 4:00pm and their bonuses are based on a metric not related to quality.

So we just have to do their job as well as our job?

Making money for the company is every employee’s job.

So the janitor should analyse and quantify what difference he will make to the companies bottom line before he wipes those skidmarks off?

Not everything is quantifiable like that. If I'm writing code and I take the time to make it 30% faster then it won't make a difference to our bottom line. If that's done everywhere then customers will like us more, be more likely to stay with us and be more likely to recommend us, but that can't be quantified in my day to day work.

Probably the biggest change looming over our industry is coding to - and deployment on - platforms that definitely are instrumented to capture exactly how much value a developer is adding (or subtracting) from the bottom line with every commit (it's a natural outgrowth of Function As A Service architectures).

It is already possible to more-or-less show how a code change (bug/fix, new feature, change in storage backend, etc.) directly affects your IAAS bill at the end of each month, and taking it a step further by tying in some instrumentation and formulas (of the sort often used to calculate the effectiveness of marketing funnels for optimization and A/B testing) to show the concomittant effect on revenue allows ROI to be calculated.

So, I am pretty sure that Finance-Oriented Programming is going to be a thing not so many years from now, tied into the whole toolchain, and very few - not developers nor their managers - seem prepared for that sea-change (some Operations folks are probably a bit better off in terms of mindset and processes).